cell adhesion molecule activity

[gocentral]

Hi

As the cell adhesion process terms are now sorted out I thought it would be easier to relaunch the request for cell adhesion molecule activity as a function term in a new request. (see https://sourceforge.net/tracker2/?func=detail&atid=440764&aid=1998775&group\_id=36855 for earlier discussions on this)

GO now has: GO:0060352, cell adhesion molecule production. GO:0050839 cell adhesion molecule binding But no terms for cell adhesion molecule activity.

In KEGG http://www.genome.ad.jp/kegg/pathway/hsa/hsa04514.html multiple CAM proteins are described. Along with http://www.genome.ad.jp/dbget-bin/www\_bget?pathway+ko04512 for ECM-receptor interaction. This details the wide variety of processes associated with CAMs and I think gives support for a GO term to describe this function.

Although the term GO:0050839 cell adhesion molecule binding would often provide sufficient description for the interaction between an ICAM and an integrin, it would not be usable to describe the interaction between CD44 and collagen, where collagen binding would be used instead. Many proteins bind collagen but not all would be described as cell adhesion molecules.

I do appreciate that the process terms would enable the description of these interactions, however if we are going to have function terms then there should be consistency. GO describes cell adhesion molecules in several terms but does not have it listed as a function in its own right.

Alex disagreed with my previous request for reinstating cell adhesion molecule activity as he felt that: 'cell adhesion molecule activity' is at heart a binding term with an unknown ligand, some structure on, presumably, another cell.

This does not seem an adequate reason to dismiss this term, there is after all GO:0060090 molecular adaptor activity which is a child of binding. Is the function of a CAM so different to this (other than being on the cell surface)?

As described by KEGG, the ligand can be on another cell, but could also be an extracellular matrix protein.

Please could the following be considered:

GO:0005515 protein binding >NTR: cell adhesion molecule activity definition: A cell surface protein interacting selectively with a cell surface protein or extracellular matrix protein. This binding activity leads to the attachment of a cell, either to another cell or to an underlying substrate such as the extracellular matrix.

Thanks

Ruth

Reported by: RLovering

Original Ticket: "geneontology/ontology-requests/5573":https://sourceforge.net/p/geneontology/ontology-requests/5573

[gocentral]

I'll leave this item open for a day or two so that people in different time zones have a chance to see it, but ultimately the ontology editors will have to reject the term.

I think Alex makes a good point about binding, but to me the more important reason not to have 'cell adhesion molecule activity' in the function ontology is that it is at heart a process term. 'Cell adhesion molecule' is simply a shorthand way of saying 'molecule involved in the process of cell adhesion'.

GO:0060090 is defined so as to take the broader phenomenon of binding and restrict it a bit based on criteria that still refer to specific aspects of the binding event.

I can think of no way to define 'cell adhesion molecule activity' that doesn't depend on invoking the process of cell adhesion.

(I'm not really sure that grouping cell adhesion molecules together as in GO:0050839 and GO:0060352 is ideal, since there are many different types of molecule involved in cell adhesion, especially if you consider yeasts, bacteria, etc.)

m

Original comment by: mah11

[gocentral]

Ruth,

I'm afraid that I have to agree with Midori. 'Cell adhesion activity' seems completely redundant with the process 'cell adhesion'. I don't think either GO:0050839 or the new GO:0060352 really establishes a precedent that relevant here, since the argument against the term is based on its simple redundancy with the process term. There's no rule saying we make X activity terms for every process X. The terms for 'cytokine activity', 'hormone activity', etc. are different in this respect in that a protein with these intrinsic activities can regulate multiple processes depending on the context.

Thanks,

Alex

Original comment by: addiehl

[gocentral]

We're concerned about the addition of new terms on the basis that there is a precedent in the ontology, since we are aware that there are existing terms in the ontology that will be better handled once column 16, see http://wiki.geneontology.org/index.php/Annotation\_Cross\_Products, is in place.

We feel that this is a slippery slope as the observation of the appearance of ANY gene product could then be included in the GO as a 'x production' term if we merely look at precedence. We added the 'cell adhesion molecule production' terms as Ruth requested because it seemed like that was the consensus that was reached after a long discussion. Neither of us was comfortable adding the terms as 'production' reflects on the measurement of a molecule rather than the biological process itself. Therefore the term describes an observation.

We allowed these types of terms for the cytokines because their appearance has specific biological effects and much of the science described in the literature infers cytokine function by measuring their production. This is not necessarily the case for cell adhesion molecules. Beta-catenin is a prime example because it can act as a cell adhesion molecule OR a signal transducer OR a transcription factor, depending on where and when it is expressed.

Tanya and David

Original comment by: tberardini

[gocentral]

I'm going to close this now, because it's clear that the ontology developers' consensus view is that we will not add/restore 'cell adhesion molecule activity' in MF. It's worth noting that we can consider making GO:0050839 and GO:0060352 obsolete when column 16 is in use.

Original comment by: mah11

[gocentral]

• status: open --> closed-rejected

Original comment by: mah11